I build multi-agent systems, LLM applications, and AI products that go beyond prototypes.
My work sits at the intersection of AI engineering, backend systems, and product development. I enjoy taking an idea from an LLM experiment to a reliable system with memory, orchestration, APIs, realtime communication, persistence, observability, and a user-facing product.
📍 Munich, Germany 📧 alam.ayaz47@gmail.com LinkedIn · Portfolio · Hugging Face
A playground for serious LLM experimentation.
A multi-LLM platform for running, comparing, and interacting with different models through a unified interface, with realtime streaming and model orchestration.
React · FastAPI · PostgreSQL · Redis · Docker · LLMs
✈️ Air Lune
What if a focus session felt like going somewhere?
Air Lune turns focused work into a shared flight.
Choose a real airport, pick a destination, invite your crew, and work together while a shared aircraft progresses across the map.
Behind the experience is a full realtime system with server-authoritative focus sessions, multiplayer presence, shared tasks, chat, browser voice, geospatial data, persistent flight history, and interactive maps.
Next.js · React · FastAPI · PostgreSQL/PostGIS · Redis · MapLibre · LiveKit · Firebase · Docker · AWS
Messages made for the right moment.
A product for creating personal messages that are meant to be opened at a specific moment, turning a simple message into a meaningful experience.
Built an enterprise multi-agent framework for Siemens Energy with persistent state, modular agent orchestration, human-in-the-loop controls, and audit-ready execution.
The system supported AI assistants for engineering knowledge retrieval, log analysis, and report generation.
Built production-grade procurement agents at Zalion GmbH using LangGraph, LangChain, FastAPI, PostgreSQL, Next.js, and AWS.
Worked on workflows including RFQ generation, supplier selection, structured request handling, validation, and human-in-the-loop execution.
Worked with 90M+ employment records at the Institute for Employment Research (IAB).
Applied XGBoost, Random Forest, SVM, and Positive-Unlabeled Learning to problems involving labor-market trends, occupational classification, skills, and job matching.
Fine-tuned GPT-2 with reinforcement learning to generate poetry.
The resulting model has reached 3,600+ downloads on Hugging Face.
Founded Ilham Labs.ai, an independent AI product studio where I build AI applications, developer tools, experiments, and product ideas.
A model is only one piece of an AI system.
The difficult and interesting engineering starts when the system needs to work reliably outside a notebook.
I care about:
- Reliable agent orchestration
- Persistent state and memory
- Human-in-the-loop systems
- Structured outputs and validation
- Retrieval and context management
- Long-running workflows
- Realtime communication
- Streaming architectures
- Observability and auditability
- Production APIs and infrastructure
- Turning AI capabilities into actual products
I like building the layer between "the model can do this" and "people can depend on this."
Python · LangGraph · LangChain · LLM Applications · Multi-Agent Systems · RAG · Prompt Engineering · Human-in-the-Loop
XGBoost · Random Forest · SVM · PU Learning · scikit-learn · SHAP · Model Evaluation
FastAPI · PostgreSQL · PostGIS · Redis · REST APIs · WebSockets · SSE · ARQ
Docker · AWS · Terraform · CI/CD · Caddy
React · Next.js · Flutter · MapLibre
LiveKit · WebSockets · Realtime Presence · Event-driven Systems
Mar 2026 to Jul 2026
Built production-grade multi-agent procurement systems using LangGraph, LangChain, FastAPI, PostgreSQL, Next.js, and AWS.
Worked across AI orchestration, structured validation, persistent workflows, backend APIs, frontend functionality, infrastructure, and CI/CD.
Jul 2025 to Feb 2026
Designed an enterprise multi-agent framework for industrial AI with persistent memory, modular orchestration, human-in-the-loop safety controls, and auditable execution.
Built AI assistants for engineering knowledge retrieval, log analysis, and report generation.
Nov 2024 to Feb 2026
Applied machine learning across 90M+ employment records, working on occupational classification, skill analysis, labor-market modeling, and job-matching prediction.
Apr 2025 to Jul 2025
Built AI agents and Generative AI applications for automated data validation, quality assessment, and enterprise productivity.
An independent AI product studio focused on building and experimenting with AI applications, LLM infrastructure, developer tools, and product ideas.
The philosophy is simple:
Build something useful. Ship it. Learn from it. Build again.
MSc Data Science Friedrich-Alexander-Universität Erlangen-Nürnberg 2022 to Present
I'm particularly interested in:
AI Engineering · Multi-Agent Systems · LLM Infrastructure · AI Products · Applied ML · Developer Tools
If you're building something where "just call an LLM" isn't enough, I'd probably enjoy working on it.
📧 Email · LinkedIn · Portfolio · Hugging Face